Table 3.

Hazard ratios (95% confidence interval) of incident diabetes for variety of fruit, vegetables, and fruit and vegetable intake in the EPIC-Norfolk Study

	Tertiles of variety of fruit intake
	Low	Medium	High	P for trend
Cases/total (n)	355/1,744	197/1,247	101/713
Mean intake (no. items/wk)	2.0 ± 1.0	4.4 ± 0.5	6.9 ± 1.2
Model 1	1 (Reference)	0.74 (0.62-0.88)	0.67 (0.54-0.84)	<0.001
Model 2	1 (Reference)	0.72 (0.60-0.86)	0.71 (0.56-0.89)	<0.001
Model 3	1 (Reference)	0.72 (0.59-0.87)	0.70 (0.53-0.91)	0.002
	Tertiles of variety of vegetable intake
Cases/total (n)	348/1,759	172/1,004	133/941
Mean intake (no. items/wk)	5.5 ± 1.4	8.5 ± 0.5	11.4 ± 1.5
Model 1	1 (Reference)	0.85 (0.71-1.02)	0.68 (0.56-0.84)	<0.001
Model 2	1 (Reference)	0.85 (0.70-1.03)	0.73 (0.59-0.89)	0.002
Model 3	1 (Reference)	0.87 (0.72-1.07)	0.77 (0.61-0.98)	0.03
	Tertiles of variety of F&V intake
Cases/total (n)	321/1,530	193/1,084	139/1,090
Mean intake (no. of items)	8.0 ± 1.8	12.0 ± 0.8	16.3 ± 2.3
Model 1	1 (Reference)	0.83 (0.69-0.99)	0.57 (0.47-0.70)	<0.001
Model 2	1 (Reference)	0.88 (0.73-1.06)	0.60 (0.49-0.74)	<0.001
Model 3	1 (Reference)	0.88 (0.73-1.07)	0.61 (0.48-0.78)	<0.001

Data are HRs (and 95% CI) estimated using Prentice-weighted Cox regression, with age as the underlying time scale variable. Adjustment for covariates was performed using multivariable Prentice-weighted Cox proportional analyses.

Model 1 was adjusted for: sex.

Model 2 as model 1 plus: BMI, waist circumference, education level, Townsend Deprivation Index, occupational social class, smoking status, physical activity, family history of diabetes, energy intake, and season

Model 3 as model 2 plus: fruit quantity for fruit variety, or vegetable quantity for vegetable variety or F&V quantity for F&V variety.